Earth and Rockfill Dams’ Seepage Prediction Using Artificial Intelligence Models: A Comprehensive Review Assessment, and Future Research Directions

Seepage prediction is vital for the safety and stability of earth and rockfill dams. Excessive seepage can lead to internal erosion and potential structural failure, thereby posing a threat to public safety and the environment. Traditional numerical and empirical models often struggle to capture the nonlinearity and complexity of seepage processes. Consequently, artificial intelligence (AI) models have emerged as a promising solution for computer-aided applications. AI models have proven to be an effective tool for improving seepage prediction accuracy in earth and rockfill dams. This review exhibited the advancements of AI applications for seepage analysis over the period (2015–2024), presenting the state-of-the-art AI models across various geological conditions. The review revealed that AI models, including classical neural networks, deep learning, ensemble models, evolutionary algorithms, and hybrid models, as well as regression methods, were employed for seepage modeling. The hybrid models demonstrated superior results in handling non-linear seepage phenomena. Additionally, deep learning models achieved effective performance in time-series forecasting, whereas kernel and dimensionality reduction models simplified the computational demands. The review revealed several challenges in the presented literature, including data availability, generalizability, and model interpretation. Future research directions were suggested from different perspectives, including the development of physics-informed AI models, data sharing, and the introduction of explainable AI models. Further, there is a need for consistent evaluation and reporting standards to guide seepage prediction using AI and facilitate reliable and reproducible practices. This review serves as a valuable guide to the potential of AI-driven seepage prediction, providing a comprehensive literature review for engineers and researchers seeking innovative solutions for dam safety management.